Proc. Natl. Acad. Sci. USAVol. 92, pp. 6289-6293, July 1995Cell Biology

Growth suppression by p16ink requires functionalretinoblastoma protein

(cell cycle/cyclin D/cyclin-dependent kinases)

RENE' H. MEDEMA, RAFAEL E. HERRERA, FELIX LAM, AND ROBERT A. WEINBERG*Whitehead Institute for Biomedical Research, Nine Cambridge Center, Cambridge, MA 02142

Contributed by Robert A. Weinberg, April 11, 1995

ABSTRACT pl6ink4 has been implicated as a tumor sup-pressor that is lost from a variety of human tumors andhuman cell lines. pl6ink4 specifically binds and inhibits thecyclin-dependent kinases 4 and 6. In vitro, these kinases canphosphorylate the product of the retinoblastoma tumor sup-pressor gene. Thus, pl6ink4 could exert its function as tumorsuppressor through inhibition of phosphorylation and func-tional inactivation of the retinoblastoma protein. Here weshow that overexpression of pl6inlk4 in certain cell types willlead to an arrest in the G, phase of the cell cycle. In addition,we show that pl6ink4 can only suppress the growth of humancells that contain functional pRB. Moreover, we have com-pared the effect of pl6ink4 expression on embryo fibroblastsfrom wild-type and RB homozygous mutant mice. Wild-typeembryo fibroblasts are inhibited by pl6ink4, whereas the RBnullizygous fibroblasts are not. These data not only show thatthe presence of pRB is crucial for growth suppression byp16ink4 but also indicate that the pRB is the critical targetacted upon by cyclin D-dependent kinases in the G1 phase ofthe cell cycle.

pl6ink4 was originally identified as a polypeptide bound tocyclin-dependent kinase 4 (cdk4) in human diploid fibroblaststransformed with the DNA tumor virus simian virus 40 (SV40)(1). The pl6ink4 gene was subsequently cloned using a two-hybrid protein interaction screen to identify proteins that canassociate with human cdk4 (2). Interestingly, the same genewas identified as a putative tumor suppressor through anextensive analysis of deletions and rearrangements of chro-mosome 9p21 present in human tumor lines (3, 4). The p16ink4gene has since been found to be deleted or rearranged in alarge number of human primary tumors (5). Binding of pl6ink4to cdk4 prevents association of cdk4 with the D-type cyclinsand results in an inhibition of the catalytic activity of the cyclinD/cdk4 enzymes (2). More recently, it was demonstrated thatp16ink4 can also bind and inhibit cdk6, the alternative kinasepartner of the D-type cyclins (6). Hence, pl6ink4 seems totarget specifically the cyclin D-dependent kinase activity bycompeting with D cyclins for binding to their kinase partners.Complexes formed by cdk4/cdk6 and cyclin D have been

strongly implicated in the control of cell proliferation duringthe G1 phase of the cell cycle (7, 8). Three different humanD-type cyclins have been identified, Dl, D2, and D3 (9, 10). Invitro, complexes of cyclin Dl, D2, or D3 and cdk4 or cdk6 canphosphorylate the product of the retinoblastoma tumor sup-pressor gene, pRB (11, 12). The phosphorylation of pRB,occurring in mid/late G1, reverses its growth-inhibitory effectand enables cells to proceed from G1 to S phase. The timingof cyclin D-dependent kinase activity and the onset of pRBhyperphosphorylation appear to coincide in the cell cycle.Taken together, these findings suggest that cyclin D/cdkcomplexes play a critical role in pRB hyperphosphorylation in

vivo. If so, pl6ink4 could negatively regulate cell proliferationby suppressing hyperphosphorylation and functional inactiva-tion of pRB.Here we show that overexpression of pl6ink4 in certain cells

will lead to an arrest in the G1 phase of the cell cycle. Incontrast, cells that lack functional pRB appear not to beaffected by high levels of pl6ink4, demonstrating that pRBmediates the growth suppression by pl6ink4. Since pl6ink4 hasbeen shown to specifically inhibit cyclin D/cdk4/cdk6 com-

plexes, these findings indicate that pRB is the critical target ofthese complexes in order to promote progression through theG, phase of the cell cycle.

MATERIALS AND METHODSPlasmid Constructs. A human pl6ink4 cDNA was obtained

by PCR amplification of a HeLa cell cDNA library (a kind giftof C. Sardet, Whitehead Institute) using two oligonucleotideprimers, a 5' primer (5'-GGAATTCACCACCATGGAGC-CTTCGGCTGAC-3') and a 3' primer (5'-GGAATTCTC-GAGTCAATCGGGGATATCTGAGGGACC-3'). A 472-bpfragment was amplified, purified on a low-melting agarose gel,and cloned directly into pGEM-T (Promega). The resultingplasmid was then used to isolate an EcoRI/Xho I fragmentcontaining the pl6ink4 coding region, which was cloned intopcDNA3 (Invitrogen) to construct pCMV.pl6ink4. pCMV.CD20, pCMV.cdk4, and pCMV.cdk2 were all provided by S.van den Heuvel (Massachusetts Institute of Technology, Cam-bridge).Recombinant Viruses. The EcoRI/Xho I fragment from

pGEM.pl6ink4 was cloned into the EcoRI/Sal I site ofpBabe.puro (13). The resulting construct was introduced intoqp-CRE cells by electroporation. Two days after electropora-tion the medium was changed to Dulbecco's modified Eaglemedium (DMEM) supplemented with 10% calf serum (CS)and 2 ,ug of puromycin per ml. Puromycin-resistant colonieswere pooled and used as producer lines. The pBabe.purovector was used to obtain a control virus-producing 4i-CREline.

Cell Culture, Transfections, and Flow Cytometry. U20S,SAOS-2, and C33A cells were cultured in DMEM supple-mented with 10% fetal calf serum (FCS). U20S and SAOS-2cells have been described previously (14); C33A cells (15) wereprovided by L. Zhu (Massachusetts General Hospital, Charles-town). 4-CRE cells (16) were cultured in DMEM supple-mented with 10% CS. Cells were transfected using thestandard calcium phosphate technique (17). Two days aftertransfection, cells were prepared for and analyzed by flowcytometry analysis as described (15).

Infections. Cultures of virus-producing tp-CRE cells were

incubated overnight with 10 ml of DMEM containing 10% CSper 10-cm tissue culture dish. The medium was filtered through

Abbreviations: cdk, cyclin-dependent kinase; RB, retinoblastoma;SV40, simian virus 40; large T, large tumor antigen.*To whom reprint requests should be addressed.

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement" inaccordance with 18 U.S.C. §1734 solely to indicate this fact.

6289

Dow

nloa

ded

by g

uest

on

Janu

ary

31, 2

020

6290 Cell Biology: Medema et al.

p2lcPl

GdG, = 72.7

S= 5.2

G2/M = 22.1

j_j.

DNA content >

FIG. 1. Gl arrest by pl6ink4. U20S osteosarcoma cells were cotransfected with 20 ,ug of the indicated plasmids and 2 ,ug ofpCMV.CD20 encodinga cell surface marker. Two days after transfection, cells were analyzed using flow cytometry. Histograms were obtained by collecting data from5000 CD20-positive cells. The cell number is plotted on they axis, and the relative DNA content is on thex axis. The percentage of cells in a particularcompartment of the cell cycle is indicated. One representative experiment of at least five independent experiments is shown.

0.22-,um filters and supplemented with 8 ,tg of Polybrene per

ml. Only freshly prepared virus stocks were used for infections.Cultures of mouse embryo fibroblasts were infected by replac-ing the medium with 5 ml of virus stock per 10-cm dish. Thisprocedure was repeated once, after which cells were culturedin DMEM with 10% FCS for 2 days. Cells were subsequentlyplaced under puromycin selection by changing the medium toDMEM with 10% FCS and 2 ,ug of puromycin per ml.

Detection of pl6ink4 Expression. Spontaneously immortal-ized mouse embryo fibroblasts infected with control or p16ink4-producing virus were metabolically labeled at 48 hr postinfec-tion with 0.1 mCi of [35S]methionine per ml (1 Ci = 37 GBq)for 4 hr. Cells were subsequently lysed in EIA lysis buffer (150mM NaCJ/50 mM Hepes, pH 7.5/5 mM EDTA/20mM NaF/1mM phenylmethylsulfonyl fluoride/10 ,tg of leupeptin perml/10 ,ug of trypsin inhibitor per ml/0.1% Nonidet P-40).pl6ink4 was immunoprecipitated with an anti-human pl6ink4polyclonal antiserum (PharMingen), and immunoprecipitateswere separated on a 15% polyacrylamide gel. For Westernblotting total cell extracts were prepared by adding 0.5 ml ofLaemmli sample buffer to a confluent 10-cm dish. Extracted

80 i

0

.,

0.

0

-5

e-

6

4

2

- \10 Id4

P., .1cu

'I

CL

Cl4

C-

_L

proteins were separated on a 15% polyacrylamide gel and blottedto nitrocellulose. p16 was detected with anti-pl6,nk4 (PharMin-gen) diluted 1:1000 in phosphate-buffered saline (PBS) to which5% dry non-fat milk was added. Bound antibody was detectedwith enhanced chemiluminescence (ECL) (Amersham) accord-ing to the manufacturer's directions.

[3H]Thymidine Incorporation. Primary cultures (passage 3)were plated on 24-well plates 1 day prior to infection. Cellswere infected with control or p16ink4-encoding virus by threesubsequent infections of 3 hr with 2 ml of conditioned mediumfrom the lp-CRE producer cells per well, supplemented with 8,Ag of Polybrene per ml. Twenty-four hours postinfection cellswere washed twice with DMEM. To each well 0.5 ml ofDMEM containing 1 ,uCi of [3H]thymidine per ml was addedand cells were labeled for 4 hr. Cells were washed with PBS andfixed with methanol. Cells were washed three times with PBSand lysed in 0.5 ml of 0.2 M NaOH. [3H]Thymidine incorpo-ration was determined by scintillation counting.

RESULTSOverexpression of pl6ink4 Causes a Gh Arrest in U20S

Ostoesarcoma Cells. A common characteristic of tumor sup-

0~-4

-A.

mC,

C-

$-

u

H HvC_, _

'.0

CL,

FIG. 2. Reversion of pl6ink4-mediated growth suppression by cdk4 and SV40 large T. (A) U20S cells were transfected with 2 ,ug of pCMV.CD20and 5 ,Ag of the pl6ink4 expression vectors as indicated, and 15 ,g of pCMV.cdk4 or pCMV.cdk2 was cotransfected as shown. (B) U20S cells weretransfected with 2 jig of pCMV.CD20, 10 j,g of pCMV.pl6ink4, and 10 ,ug of pCMV.LT (large T) as indicated. The total amount of DNA in eachtransfection was normalized to 22 ,ug by addition of pcDNA3. Two days after transfection, cells were analyzed using flow cytometry. Black barsrepresent the percentage of GO/Gl-phase cells, white bars represent S-phase cells, and gray bars represent G2/M-phase cells. Standard errorscalculated from percentages obtained from three independent experiments are indicated.

control

GOIG, = 33.0S = 32.6

G2/M = 34.5

pl6ink4

GO/G = 68.0S = 9.3

G2/M = 22.7

A

~0

10

0:

Proc. Natl. Acad. Sci. USA 92 (1995)

Dow

nloa

ded

by g

uest

on

Janu

ary

31, 2

020

Proc. Natl. Acad. Sci. USA 92 (1995) 6291

pressor genes is their ability to inhibit cell proliferation whenoverexpressed in sensitive cell lines (14, 15, 18, 19). Todetermine whether pi6ink4 can inhibit cell proliferation, weemployed a modified flow cytometry technique (15). Controlvectors or expression vectors coding for pl6ink4 were cotrans-fected with a plasmid expressing the CD20 cell surface marker(20). Cotransfection of this plasmid enabled us to identify cellsthat contain transfected DNA by staining with an anti-CD20monoclonal antibody conjugated to fluorescein isothiocya-nate, and the DNA content of the transfected cells could bemeasured independently by propidium iodide staining.

Transfection of an expression plasmid encoding pl6ink4 inthe human osteos4rcoma cell line U20S resulted in a verysignificant increase (-35%) in the fraction of CD20-positivecells in the Go/G1 phases of the cell cycle, consistent with itsputative role as tumor suppressor. Similarly, transfection ofanother inhibitor of cyclin/cdks, p2lciPl, also caused an accu-mulation of cells in the Go/GI phases of the cell cycle (Fig. 1).This latter inhibitor is known to antagonize several cyclin/cdkcomplexes, including cdks 2, 4, and 6 (21-23), and our ownwork has confirmed that overexpression of p21ciPl in a varietyof cell lines causes a G, arrest (R.H.M. and R.A.W., unpub-lished results).Growth Suppression by pl6ink4 Is Abolished by cdk4 and

SV40 Large Tumor Antigen (Large T). Since pl6ink4 competeswith cyclin D for binding to cdk4 and cdk6, we reasoned thatsubstantial overexpression of cdk4 should partially or com-pletely relieve the p16ink4-imposed cell cycle arrest. Indeed, theincrease in the Go/G1 population established by pl6ink4 inU20S cells was greatly reduced when a plasmid coding forhuman cdk4 was cotransfected (Fig. 2A). In contrast, cotrans-fection of an expression vector encoding human cdk2 did notsignificantly affect the arrest mediated by pl6ink4 (Fig. 2A).This reinforces the notion that the inhibition of cell prolifer-ation by pl6ink4 is mediated through its ability to bind cdk4/6and indicates that pl6ink4 acts as an inhibitor of cyclin D/cdkcomplexes not only in vitro but also in vivo.Upon transformation with a variety of viral oncoproteins,

cdk4 is no longer found in complex with D cyclins butassociates exclusively with pl6ink4 (1). This finding suggeststhat the critical substrates of cyclin D/cdk complexes them-selves have been inactivated by viral transformation, therebyobviating cyclin D/cdk activity. To confirm that cyclin D/cdk4/cdk6 function is no longer needed in the presence ofDNA tumorvirus oncoproteins, we cotransfected a SV40 large T expressionvector together with a pl6ink4 vector. Fig. 2B shows that growthinhibition induced by pl6ink4 was completely reversed by cotrans-fection of SV40 large T. This demonstrates that SV40 large T canindeed drive cell proliferation in cells that lack cyclin D-dependent kinase activity. Similarly, it was recently demonstratedthat suppression of cellular transformation by pl6ink4 can beovercome by the ElA oncoprotein (24). Since SV40 large T andElA are known to bind and inactivate pRB, these findings seemconsistent with the hypothesis that pl6ink4 inhibits cell prolifer-ation through its effects on pRB. However, based on theseobservations one cannot rule out that growth suppression by

pl6ink4 might be mediated through other viral oncoprotein tar-gets-for example, p107 or p130, both of which are also boundand inactivated by SV40 large T and ElA.

Overexpression of pl6ink4 Does Not Arrest Human CellLines That Lack Functional pRB. To resolve among thesepossibilities, we studied the effects of overexpressing p16ink4 intwo human tumor cell lines known to lack functional pRB,specifically the SAOS-2 osteosarcoma cell line and a cervicalcarcinoma cell line, C33A. Expression of p16ink4 in these celllines did not give rise to any increase in the G, population, insharp contrast to the G1 arrest previously observed in U20Scells (Table 1). However, expression of p2lciPl blockedSAOS-2 and C33A cells in G1, similar to the effect it had inU20S cells (Table 1). This is in agreement with the obser-vation that p2lciPl inhibits a larger array of cyclin/cdk com-plexes than does p1oink4 and inhibits DNA replication by amechanism independent of cyclin/cdks (25). A similar lack ofgrowth suppression by p16ink4 when expressed in human celllines that do not contain functional pRB was recently observedby others (24, 26). However, while consistent with the role ofpRB as the critical mediator of cyclin D/cdk4/6 control, theseobservations are hardly definitive, since many distinct geneticlesions are likely to exist in these cell lines. Moreover, SAOS-2cells have been reported to have very low levels of cyclin Dl(14, 27, 28), and complexes of Dl and cdk4 or cdk6 could notbe detected in C33A cells (27, 28), further confounding theinterpretation of these data.

Effects of pl6ink4 on Wild-Type and RB Mutant MouseEmbryo Fibroblasts. We reasoned that unequivocal evidencefor a role for pRB in p16ink4_mediated growth arrest could onlybe obtained through a comparison of two cell types that aregenetically identical, except for the presence or absence ofpRB. For this reason, we decided to study the effects ofoverexpression of p16ink4 in primary mouse embryo fibroblastsderived from wild-type embryos and from RB nullizygousembryos (29). Another advantage of using these cells is thatthe endogenous levels of cyclin Dl and cdk4 are very similarin both cell types (R.E.H., unpublished observations). Toexpress p16ink4 in these cells, we generated a retrovirus car-rying the p16ink4 coding region and the puromycin-resistancegene, the latter being used to select for successfully infectedcells. As a control, we infected cells with a similar virus thatlacked the p16ink4 coding region.

Infection of RB nullizygous fibroblasts as well as the wild-type fibroblasts with the control virus gave rise to a large,similar number of puromycin-resistant cells, confirming thecomparable infectability of the two cell types (Fig. 3). How-ever, infection of these two cell types with the much lower-titered p16ink4 virus gave a very different outcome: 10-20colonies per 10-cm culture dish were obtained upon infectionof the pRB homozygous mutant cells, while no colonies arosefrom the infected wild-type cultures (Fig. 3). This providedclear evidence that the presence of pRB is critical to p16ink4_mediated growth arrest.To substantiate further these observations, we determined

the levels of p16ink4 following infection with the p16ink4 retro-

Table 1. Absence of a p16ink4-induced growth arrest in human tumor lines that lack functional pRB

Transfected plasmid

Control pCMV.p2lciPl pCMV.pl6ink4Cell line Go/GI S G2/M Go/G1 S G2/M Go/Gi S G2/MU20S 44+4 19±4 37±5 73+4 5±4 22±5 68+6 9±5 23±2SAOS-2 47 2 22 ± 2 31 ± 4 64 1 9 ± 2 27 ± 1 36 ± 1 21 ± 1 43 ± 1C33A 48 4 24 ± 4 28 ± 5 63 4 14 ± 4 23 ± 5 44 ± 6 22 ± 5 34 ± 2

Cells were transfected with 20 jig of the indicated plasmid and 2 ,ug of pCMV.CD20. Two days after transfection the cellswere prepared for flow cytometry. Indicated is the percentage of CD20-positive cells present in the respective phases of thecell cycle. The percentages and standard errors shown here are obtained by combining the data of three independentexperiments.

Cell Biology: Medema et aL

Dow

nloa

ded

by g

uest

on

Janu

ary

31, 2

020

6292 Cell Biology: Medema et al.

Mock Control p16ink4

t.e ^,. ..^ .. .... ..X . , ., i.w .tw ..... ....f W..t,....

=!<<E z23m. vs.-w ;'s>.s..; '-s.;ES,N }4; m MS.:; -3 P>;" 1B,G,,.M ,.D D _ / ' .t,,li z {Suz . #: . ,jj, t, :

D/.R,;S^, {, ''. ,&.ffiiV:',g, r. ,,it4t;¢,sji rv.

.<^'t.$8::s.ssrs:t: '..''.': :.'.3fi.!.f,,:',,.>, t _, 3: .. .;: ...::.. ;' "

FIG. 3. Effect of p16ink4 on wild-type and RB nullizygous mouse embryo fibroblasts. RB homozygous mutant (Lower) and wild-type (Upper)embryo fibroblasts were infected with pl6ink4-encoding virus or control virus as indicated. Mock-infected cells were maintained in DMEMcontaining 10% CS for the same period of time. Twenty-four hours postinfection the growth medium was switched to medium containing 2 ,ugof puromycin per ml. Pictures were taken 2 weeks after initiation of puromycin selection. At this time the uninfected dishes no longer containedany cells. (x40.)

virus vector. Since the number of successfucells was rather low, we infected spontaneRB nullizygous and wild-type embryo fp16ink4-encoding virus. As shown in Fig. 4,amounts of pI6ink4 could indeed be immuiinfection with this virus (in this case fror(lane 2). In addition, a clear coimmunopseen (Fig. 4A) that comigrates with murinUpon puromycin selection of these imiinfected with the p16ink4 virus, a large Jnullizygous fibroblasts continued to grow10-cm dish), but only a very small numbesurvived (2-5 colonies per 10-cm dish). We

A

kDa43 -

B

1 2

illy infected primary of p16ink4 in these immortalized, puromycin-selected wild-type-ously immortalized and RB mutant populations by immunoblot analysis. pl6ink4.ibroblasts with the was readily detected in the RB nullizygous population (Fig. 4B,A, greatly increased lane 2), while no p16ink4 was detectable in the wild-type cellsnoprecipitated after that survived infection with the p16ink4 virus followed bym the mutant cells) puromycin selection (Fig. 4B, lane 3). This clearly indicatestrecipitating band is that the RB nullizygous cells can tolerate high levels of p16ink4,ie cdk4 (not shown). whereas the wild-type cells cannot.mortalized cultures Finally, we investigated whether infection with the p16ink4fraction of the RB virus would cause a significant inhibition of S-phase entry. To(>105 colonies per do so, we infected primary cultures of wild-type and RB-r of wild-type cells homozygous fibroblasts with high multiplicities of the controlcompared the level and p16ink4 viruses and measured levels of [3H]thymidine

incorporation 24 hr after infection (Fig. 5). A very significantinhibition (44%) of thymidine incorporation was observed inwild-type embryo fibroblasts infected with the p16ink4 viruscompared to the control virus. In contrast, the fibroblasts fromthe RB nullizygous embryos were unaffected. From this ex-periment we cannot distinguish whether the observed reduc-

1 2 3 tion in thymidine incorporation in the wild-type cells is the29 -

100

18.4 - 00

0

75

50

14.3 -

FIG. 4. Expression of pl6ink4 in spontaneously immortalizedmouse embryo fibroblasts. (A) Spontaneously immortalized fibro-blasts derived from RB nullizygous mouse embryos were infected withcontrol (lane 1) or pl6ink4 encoding virus (lane 2). Two days postin-fection cells were metabolically labeled, pl6ink4 was immunoprecipi-tated, and immunoprecipitates were separated on a 15% polyacryl-amide gel. Molecular size markers are shown in kDa. The coimmu-noprecipitating band of 32 kDa is indicated with an arrow. (B)Spontaneously immortalized fibroblasts derived from RB nullizygous(lanes 1 and 2) or wild-type (lane 3) mouse embryos were infected withcontrol (lane 1) or p16ink4-encoding virus (lanes 2 and 3). Two dayspostinfection the culture medium was changed to medium containing2 ,ug of puromycin per ml. Cells were grown on this selective mediumfor 2 months, after which the amount of pl6ink4 in each population wasdetermined by Western blotting. Equal amounts of protein wereloaded in each lane.

25

0 ~ 0

0 0

+1/+ _-I-FIG. 5. pl6ink4 blocks S-phase entry in wild-type but not RB

nullizygous fibroblasts. Primary cultures of wild-type and RB nullizy-gous mouse embryo fibroblasts were seeded on 24-well plates andinfected with control or p16ink4-producing virus. Twenty-four hoursafter infection [3H]thymidine incorporation was determined. Values ofthymidine incorporation from the cultures infected with the controlvirus were set at 100%.

Proc. Natl. Acad. Sci. USA 92 (1995)

,-. -.- I pl6lNK4A ->--

Dow

nloa

ded

by g

uest

on

Janu

ary

31, 2

020

Proc. Natl. Acad. Sci. USA 92 (1995) 6293

result of an arrest in G, or in another phase of the cell cycle.However, it clearly demonstrates that the effect of pl6ink4 oncell cycle progression is mediated by pRB and seems mostconsistent with an arrest in GI.

DISCUSSIONHere we show that the growth-suppressive effect of pl6ink4 isdependent on the presence of functional pRB. Since theactions of pl6ink4 are likely mediated largely, if not entirely,through inhibition of cyclin D/cdk4/cdk6 complexes, thesefindings indicate that pRB is the critical target of thesecomplexes. These findings are consistent with the observationthat microinjection of anti-cyclin Dl antibodies or cyclin Dlantisense plasmids, which causes a G1 arrest in normal fibro-blasts, has no effect in cells lacking functional pRB (30). Incontrast, microinjection of anti-cyclin E antibodies will causea G1 arrest in wild-type as well as RB mutant cells, indicatingthat this latter cyclin has distinct substrates that are critical forcell cycle progression (31).The evidence presented here connects two tumor suppres-

sors, p16 inl4 and pRB, in one pathway controlling cell growth.In this pathway cyclin D/cdk complexes would mediate. theinactivation of pRB, either directly or indirectly. This inacti-vation causes the release of several proteins, including the E2Ftranscription factors, which allows entry into S phase. Inter-estingly, a number of distinct genetic aberrations have beenobserved in human tumors, which could all possibly result inloss of growth regulation through this pathway. On the onehand, loss of pRB or p16inlk4 by point mutation or deletion hasbeen demonstrated in a variety of human tumors (3, 4, 32). Onthe other hand, translocation and amplification of cyclin Dl orcdk4 have also been observed in a number of tumors (33-36).Finally, although no examples are known at present of ampli-fication of one of the E2F family members in human tumors,overexpression of E2F-1 was shown to cause neoplastic trans-formation of cells in culture (37).

Several lines of evidence suggest that inactivation of thisgrowth-regulatory pathway might be an absolute requirementfor the progression of certain tumors. A recent report dem-onstrated that 85% of the human astrocytomas analyzed showedeither homozygous deletion of pl6ink4 or, altematively, amplifi-cation of the cdk4 gene (38). Also, a number of groups recentlyobserved an inverse correlation between the expression of wild-type pRB and pl6fink4 in many human tumor cell lines (39-41).Now that it is becoming evident what the components of thispathway are, it will be of interest to check the status of each ofthese components within one type of tumor.

We thank the other members of the Weinberg laboratory, inparticular Tomi Makela, for critical discussions and help with the workdescribed in this manuscript; Claude Sardet for critically reading themanuscript; Tyler Jacks, Bart Williams, and George Mulligan forproviding the primary embryo fibroblasts used in this study; Sandervan den Heuvel for the pCMV.CD20, pCMV.cdk2, and pCMV.cdk4expression plasmids; and Liang Zhu for the C33A cell line. R.H.M. wassupported by a postdoctoral fellowship from the Dutch Cancer Society.R.A.W. is a Research Professor of the American Cancer Society andreceives support from the National Cancer Institute.

1. Xiong, Y., Zhang, H. & Beach, D. (1993) Genes Dev. 7, 1572-1583.

2. Serrano, M., Hannon, G. J. & Beach, D. (1993) Nature (London)366, 704-707.

3. Kamb, A., Gruis, N. A., Weaver-Feldhaus, J., Liu, Q., Harshman,K., Tavtigian, S. V., Stockert, E., Day, R. S., III, Johnson, B. E.& Skolnick, M. H. (1994) Science 264, 436-440.

4. Nobori, T., Miura, K., Wu, D. J., Lois, A., Takabayashi, K. &Carson, D. A. (1994) Nature (London) 368, 753-756.

5. Sheaff, R. J. & Roberts, J. M. (1995) Curr. Biol. 5, 28-31.6. Hannon, G. J. & Beach, D. (1994) Nature (London) 371,257-261.7. Sherr, C. J. (1993) Cell 73, 1059-1065.8. Hinds, P. W., Dowdy, S. F., Eaton, E. N., Arnold, A. & Weinberg,

R. A. (1994) Proc. Natl. Acad. Sci. USA 91, 709-713.9. Inaba, T., Matsushime, H., Valentine, M., Roussel, M. F., Sherr,

C. J. & Look, A. T. (1992) Genomics 13, 565-574.10. Xiong, Y., Menninger, J., Beach, D. & Ward, D. C. (1992)

Genomics 13, 575-584.11. Kato, J., Matsushime, H., Hiebert, S. W., Ewen, M. E. & Sherr,

C. J. (1993) Genes Dev. 7, 331-342.12. Meyerson, M. & Harlow, E. (1994) Mol. Cell. Biol. 14,2077-2086.13. Morgenstern, J. P. & Land, H. (1990) Nucleic Acids Res. 18,

3587-3596.14. Hinds, P. W., Mittnacht, S., Dulic, V., Arnold, A., Reed, S. I. &

Weinberg, R. A. (1992) Cell 70, 993-1006.15. Zhu, L., van den Heuvel, S., Helin, K., Fattaey, A., Ewen, M.,

Livingston, D., Dyson, N. & Harlow, E. (1993) Genes Dev. 7,1111-1125.

16. Danos, 0. & Mulligan, R. C. (1988) Proc. Natl. Acad. Sci. USA85, 6460-6464.

17. van der Eb, A. J. & Graham, F. L. (1980) Methods Enzymol. 65,826-839.

18. Baker, S. J., Markowitz, S., Fearon, E. R., Willson, J. K. &Vogelstein, B. (1990) Science 249, 912-915.

19. Mercer, W. E., Amin, M., Sauve, G. J., Appella, E., Ullrich, S. J.& Romano, J. W. (1990) Oncogene 5, 973-980.

20. Stamenkovic, I. & Seed, B. (1988) J. Exp. Med. 167, 1975-1980.21. Gu, Y., Turck, C. W. & Morgan, D. 0. (1993) Nature (London)

366, 707-710.22. Harper, J. W., Adami, G. R., Wei, N., Keyomarsi, K. & Elledge,

S. J. (1993) Cell 75, 805-816.23. Xiong, Y., Hannon, G. J., Zhang, H., Casso, D., Kobayashi, R. &

Beach, D. (1993) Nature (London) 366, 701-704.24. Serrano, M., Gomez-Lahoz, E., DePinho, R. A., Beach, D. &

Bar-Sagi, D. (1995) Science 267, 249-252.25. Waga, S., Hannon, G. J., Ejeach, D. & Stillman, B. (1994) Nature

(London) 369, 574-578.26. Guan, K. L., Jenkins, C. W., Li, Y., Nichols, M. A., Wu, X.,

O'Keefe, C. L., Matera, A. G. & Xiong, Y. (1994) Genes Dev. 8,2939-2952.

27. Bates, S., Parry, D., Bonetta, L., Vousden, K., Dickson, C. &Peters, G. (1994) Oncogene 9, 1633-1640.

28. Lukas, J., Pagano, M., Staskova, Z., Draetta, G. & Bartek, J.(1994) Oncogene 9, 707-718.

29. Jacks, T., Fazeli, A., Schmitt, E. M., Bronson, R. T., Goodell,M. A. & Weinberg, R. A. (1992) Nature (London) 359, 295-300.

30. Tam, S. W., Theodoras, A. M., Shay, J. W., Draetta, G. F. &Pagano, M. (1994) Oncogene 9, 2663-2674.

31. Ohtsubo, M., Theodoras, A. M., Schumacher, J., Roberts, J. &Pagano, M. (1995) Mol. Cell. Biol. 15, 2612-2624.

32. Weinberg, R. A. (1992) Science 254, 1138-1146.33. Motokura, T., Bloom, T., Kim, H. G., Juppner, H., Ruderman,

J. V., Kronenberg, H. M. & Arnold, A. (1991) Nature (London)350, 512-515.

34. Withers, D. A., Harvey, R. C., Faust, J. B., Melnyk, O., Carey, K.& Meeker, T. C. (1991) Mol. Cell. Biol. 11, 4846-4853.

35. Reifenberger, G., Reifenberger, J., Ichimura, K., Meltzer, P. S. &Collins, V. P. (1994) Cancer Res. 54, 4299-4303.

36. Khatib, Z. A., Matsushime, H., Valentine, M., Shapiro, D. N.,Sherr, C. J. & Look, A. T. (1993) Cancer Res. 53, 5535-5541.

37. Singh, P., Wong, S. H. & Hong, W. (1994) EMBO J. 13, 3329-3338.

38. Schmidt, E. E., Ichimura, K., Reifenberger, G. & Collins, V. P.(1994) Cancer Res. 54, 6321-6324.

39. Otterson, G. A., Kratzke, R. A., Coxon, A., Kim, Y. W. & Kaye,F. J. (1994) Oncogene 9, 3375-3378.

40. Okamoto, A., Demetrick, D. J., Spillare, E. A., Hagiwara, K.,Perwez Hussain, S., Bennett, W. P., Forrester, K., Gerwin, B.,Serrano, M., Beach, D. H. & Harris, C. C. (1994) Proc. Natl.Acad. Sci. USA 91,' 11045-11049.

41. Tam, S., Shay, J. & Pagano, M. (1994) CancerRes. 54,5816-5829.

Cell Biology: Medema et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

31, 2

020